As the development of treatments for Alzheimer's disease (AD) continues, there is an urgent need for biomarkers that can diagnose AD as early as possible, and that can map disease progression for the purposes of testing drug efficacy. Previous studies have suggested that there is a strong link between resting cerebral blood flow (CBF) assessed by PET or SPECT and neuropathological changes of AD. However PET or SPECT are limited by factors such as high cost, low availability, invasive nature, ionizing radiation or low spatial resolution. Arterial spin labeling (ASL) MRI is a relatively recent, noninvasive, and cost-effective imaging technique which provides absolute quantification of CBF with reproducibility, resolution and contrast exceeding that obtained with PET or SPECT. This proposal combines the measuring power of ASL MRI with the analytic power of multivariate analysis methods. Despite the attractive features of multivariate analytic techniques compared with more commonly used univariate techniques, they have rarely been used to study the neural correlates of AD or cognitive impairment. Multivariate analysis methods can identify patterns of regionally correlated CBF changes that might precede the clinical and structural manifestation of the disease, and therefore serve as sensitive early diagnostic tool. 20 healthy elderly controls, 60 subject with Mild Cognitive Impairment, and 20 early AD subjects will be scanned repeatedly with ASL MRI during rest to establish CBF patterns underlying early AD. Their ability to detect AD, predict disease progression and predict of conversion to AD status in the MCI subjects will be investigated. It is feasible that CBF measurement through ASL and forward application of such AD-related CBF patterns can become a routine part of structural MRI scans, to capture in one number -similar to a neuropsychological measure- the degree to which subjects display AD-related CBF changes in at-risk and early stage patients, and thereby aid the early detection and treatment. [unreadable] [unreadable] [unreadable]